-- 예시자료와 결과자료가 일치하지 않네요..임의로 조정해서 해봤습니다.
WITH t AS
(
SELECT LEVEL lv
FROM dual
WHERE LEVEL NOT IN (7, 9, 15, 22, 23, 24, 25, 33, 36)
CONNECT BY LEVEL <= 40
)
SELECT MIN(DECODE(MOD(rn, 4), 1, v)) v1
     , MIN(DECODE(MOD(rn, 4), 2, v)) v2
     , MIN(DECODE(MOD(rn, 4), 3, v)) v3
     , MIN(DECODE(MOD(rn, 4), 0, v)) v4
  FROM (SELECT ROW_NUMBER() OVER(ORDER BY MIN(lv)) rn
             , MIN(lv) || DECODE(COUNT(*), 1, '', '~'||MAX(lv)) v
          FROM (SELECT lv
                     , ROWNUM rn
                     , COUNT(*) OVER(PARTITION BY ROWNUM - lv) cnt
                  FROM (SELECT lv FROM t ORDER BY lv)
                )
         GROUP BY lv - rn
             , CASE WHEN cnt < 5 THEN lv END  
        )
 GROUP BY CEIL(rn / 4)
 ORDER BY CEIL(rn / 4)
;

----------------------------
V1     V2     V3     V4
----------------------------
1~6    8      10~14  16~21
26~32  34     35     37
38     39     40
----------------------------

with t as
( select distinct ceil(level+dbms_random.value(1, 5)) no
 from dual connect by level <= 30
)
select wm_concat(val) keep(dense_rank first partition by ceil(rownum/4)) val
from (
   select min(no) no
      , decode(sign(min(no)-max(no)), 0, to_char(min(no)), min(no)||'~'||max(no)) val
   from (select no 
         , no - row_number() over(order by no) val
      from t)
   group by val
   order by no
)
group by ceil(rownum/4)
order by ceil(rownum/4);